Software ate cleantech.

Now what?

Today, my colleagues Varun Sivaram and Frank O’Sullivan and I released a report through the MIT Energy Initiative on the past, present, and future of cleantech venture capital.

As we wrote this morning in an op/ed in the Financial Times:

In 2006, Silicon Valley began to bet big on clean energy technology. Seduced by grand visions of making a fortune while saving the planet, venture capitalists invested a then-record $123m in the first round of fundraising for 16 new companies that year. In 2008, they would sink nearly $1bn in over 100 new companies.

But when these investments began to flop, the cleantech bubble abruptly popped. Since 2009, VCs have barely funded 25 new cleantech companies a year, slowing new investment to a trickle.

What went wrong? And where should cleantech go from here? To answer these questions, we compared the performance of every medical technology, software technology and cleantech company that received its first round of VC funding between 2006 and 2011. We found that betting on cleantech start-ups just did not make sense for VCs, because cleantech could not deliver the outsized returns found in other sectors.

This conclusion is alarming because new technologies are desperately needed to confront climate change. Still, guided by the lessons learnt from the cleantech VC boom and bust, new private and public funding sources may be able to better support revolutionary technologies.

Challenging the conventional wisdom: Cleantech just needs more money and more time

We hope that our findings will dispel the notion that the climate change can be solved by simply replacing VC money with patient capital. While it is true that the R&D needed to bring breakthrough innovations to market often takes decades and that building factories and manufacturing products requires substantial capital, there are further lessons to be learned from the first cleantech bubble.

What went wrong?

In short, investments in cleantech — especially those in companies commercializing breakthrough energy technologies — simply couldn’t deliver the outsized returns required by the venture capital model. We found five basic reasons that these companies failed:

  1. Long development cycles. Developing breakthrough materials, such as those for next generation solar panels, can take 20-30 years.
  2. Significant capital requirements. To reach scale, venture capital money was used to build factories, in some cases before R&D was done.
  3. Business model failures. Many companies faced expensive customer acquisition and long sales cycles.
  4. No sales premium. Energy companies were selling into established commodity markets where price (of electricity or of the solar panel itself) was the main driver.
  5. Few exit opportunities. Acquirers in the utility and industrial sectors were not eager to take risks or place a premium on a startup’s growth.

Software ate cleantech.

The result was an overall retrenchment of venture capital in cleantech companies. Total funding for startups in the sector fell from $5 billion in 2008 to $2 billion in 2012. More importantly, funding for new companies took an even harder hit. In 2008, nearly 120 companies received their first major round of funding. In 2009, fewer than 30 companies could say the same — a number that has remained approximately constant every year since.

What about the deals that are still getting done?

VCs all but stopped funding “deep technology” companies and shifted their focus largely towards software.


A-round VC financing for cleantech companies shifted from breakthrough materials, chemicals, and hardware to “capital light” innovations and business models.

Who will fund the breakthroughs?

In order to avoid the worst effects of climate change through the year 2050 and beyond, we absolutely must both deploy as much existing technology as we can, as quickly as we can manage, and with clever new business models — but the magnitude of the problem will require new breakthroughs. There has been a substantial evolution of the ecosystem since the boom and bust, but the battle is not won yet.

First, public-sector support of energy R&D has been buoyed by ARPA-E, the Department of Energy’s advanced research projects arm. More recently, in committing to Mission Innovation at the Paris climate talks last year, President Obama pledged to double the nation’s energy R&D budget over the next several years. Other commercialization and technology transfer efforts within the Department of Energy have provided increased access to our National Laboratories and infused an entrepreneurial culture inside the Lab system. In just the past few years, DOE has launched Cyclotron Road and Chain Reaction Innovations, Small Business Vouchers, and LabCorps, and has established the new Office of Technology Transitions.

New sources of capital are also available. A number of foundations, family wealth management offices, and high net-worth individuals have begun pledging to support these breakthrough technologies. Keep in mind that these are largely not new actors: these investors were the source of capital originally deployed by the VCs in the first wave of cleantech. Now, however, some feel that by investing directly or through syndicates, they may avoid the short term horizon imposed by the typical venture structure. Last Fall, Bill Gates and 27 other billionaires announced the Breakthrough Energy Coalition, pledging to put billions of dollars to work over the long term to support emerging technologies. Meanwhile, the PRIME coalition is helping foundations and philanthropists invest directly in game-changing technology companies, often before other institutional investors would be willing to take such risks.

But capital is only part of the equation. Cleantech companies require much of the same support as other types startups: help with developing sales strategies, hiring the right key employees, marketing, and fundraising. But they also require some things a SaaS company doesn’t: lab space, manufacturing expertise, and — particularly important when selling to risk-averse adopters — opportunities to test and demonstrate their technology. In the decade since the cleantech boom started, a number of incubators and accelerators have stepped in to fill these gaps. My organization, Clean Energy Trust, invests in and works directly with startups to help them through this early scale-up process. Others, such as the Los Angeles Cleantech Incubator and Greentown Labs in Somerville, MA provide physical incubation space for new companies. Cleantech companies now have improved access to first-of-a-kind demonstration and pilot projects through programs like the Wells Fargo/NREL Innovation Incubator, the Energy Excelerator demonstration track, and CET’s Campus Cleantech Pilots. Large industrial corporations and utilities are also increasingly becoming aware of the need to work with innovative companies. For example, through our Cleantech Innovation Bridge we match startups with corporate partners in need of emerging technologies.

No Exit: Funding the slow-growth company

There is no doubt that the support structures for new emerging technology companies has gotten stronger and we hope that this will fuel increased interest in solving climate change through innovation. Before 2012, some cleantech companies were able to raise over $500 million in venture capital before turning a profit. That is simply not likely to happen again.

The model for success in the current funding environment has not yet been proven, though there are a few predictions we can make. To succeed, cleantech founders will have to run lean. First, they will have to take advantage of incubators and accelerators and will need to fund R&D through public grant funding. Later, they will need to rely on debt for projects (such as Black Coral and Generate Capital) and manufacturing (including private working capital revolving funds and public sources such as OPIC) as an alternative to equity financing. Throughout the process, cleantech founders should be in close contact with corporate and strategic partners who will not only be the likely customer for the product, but also potential acquirers of the company.

Finally, cleantech companies will need to build a business on the assumption that they will be valued based on their profits, not their growth.

Closing thoughts

Venture capital didn’t work for cleantech because there were better places to invest. That doesn’t mean it’s not a sector worth investing in. As many others have put it: climate change is the greatest challenge we have faced in a generation and the transition to a low-carbon energy system will be the single greatest economic opportunity in our lifetimes.

Further reading:

Op/Ed in the Financial Times

Working paper at the MIT Energy Initiative

Full article at SSRN

The end of the cleantech leasing model?

Will your clever cleantech business model bankrupt your company?

If you spend enough time talking to people in cleantech, you’ll often hear them talk excitedly about how new business models will be key to unlocking the low-carbon energy economy. But what if it could end up killing your company?

Let’s start with the basics: a business model is “all the parts of a strategy necessary to deliver a product to a customer and make money from it,” as Steve Blank describes it. As Blank tells it, this includes “the product itself, the customer, the distribution channel, the revenue model, how to get, keep and grow customers, resources and activities needed to build the business, and costs.”

In world of Silicon Valley-funded software startups, the subscription model where customers make recurring payments (software as a service, or SaaS) has come to dominate, even if it’s not a new idea. For software companies, this has been enabled by relatively cheap cloud computing. But other benefits such as increased customer retention (“stickiness”) have helped justify the old-fashioned magazine-and-newspaper subscription model in other industries–think razors, snacks, clothes, and even dog toys.

In cleantech, business model innovation often means a different revenue model. The first version of this was a leasing model, which gives customers the advantage of low up-front costs and gives the startup predictable cash flows. Think of a company that, instead of selling solar panels, installs the panels on a customer’s roof, maintains ownership of the panels, and simply sells electric power. There are some very inventive new revenue models that take advantage of “ancillary markets” which are in essence things a utility will pay you to do or not do. For instance, a smart battery system in your garage can communicate with the utility to help regulate the frequency—and the utility will pay you for that service. Some new startups are building businesses around these multiple revenue streams, and offering the battery hardware for free, up front, with recurring payments based on these new revenues. This is a very clever variation on a lease, and while it’s tempting to compare the leasing model to SaaS and consumer goods subscriptions because all three models yield monthly cash flows, the reality is that the differences make leasing a tough route for new startups.


Cleantech leasing models aren’t a new concept. (You could even argue that electric utilities have been operating a kind of leasing model since the 1880s.) In the early 2000s, Jigar Shah’s SunEdison LLC began developing solar projects where they provided the up-front capital for the equipment and installation, then sold electricity via Power Purchase Agreements (PPAs). This meant that a customer who was interested in solar didn’t have to worry about buying, installing, or maintaining the solar power plant, but still got to enjoy the benefits. This model meant that not only did the customer avoid worrying about the risk of the new technology (if the panels broke, the customer wasn’t on the hook to fix them) but also let them avoid the hefty up-front investment.

In 2008, SolarCity applied the model to residential rooftop solar. A homeowner could have solar panels installed on their roof with no money down, and simply pay a monthly fee. Other solar companies quickly followed suit (including Sungevity, BrightGrid, and SunRun) and the model was ported to other technologies (LEDs from Lemmis Lighting and Green Ray, fuel cells from Bloom, electric vehicle chargers from ChargePoint, batteries from Stem, as just a few examples).

Problems with the No Money Down Model

“No money down” is very appealing to customers, especially for energy efficiency technology, so it’s not surprising that so many cleantech startups raising money today pitch some sort of recurring revenue or subscription model. Of course, there are cleantech software businesses that might be a great fit for the model, but let’s focus on the companies selling tangible energy products.

There are two interrelated problems with the leasing/subscription/cleantech-as-a-service model. The first problem is one of mismatched growth expectations and the second is that these companies will run out of cash, fast.

Problem 1: Growth

When you read that a SaaS startup is valued at 10x ARR (annual recurring revenue), it’s easy to run the numbers and try to guess what that means for your company. But here’s the key difference: investors in the private and public markets place a value on subscription businesses based on their revenue and their growth. Subscription companies grow by spending huge amounts of money on sales, and they can afford to do this because their product is scalable. Once the software is written and sold to the first customer, the cost of delivering the software to the second customer is very small. Take a company like SalesForce, an industry-leading SaaS company, as an example. For every $1 of revenue, SalesForce spends about $0.50 on sales, $0.25 getting the software to the customer, $0.15 on writing the software. The sad reality is that making physical products is more expensive. For every $1 of Tesla’s revenue, $0.70 goes into making the product. (And for those who are counting: at Ford the number is $0.77).

This model works well for software startups. Acquirers are willing to pay for that future growth. Over the last decade, we’ve seen the acquirers in the cleantech space are not willing to pay that growth premium, and instead focus on profitability.

Problem 2: Cashflow

That brings us to the second problem: running out of cash.

Let’s say your company is selling solar panels. You have a bit of seed money in the bank, so you use that to manufacture your product and then go out and make a sale. You take the money your customer just paid you and plow the profits back into making more panels.

Suppose instead you adopt the leasing model. You take your seed capital, manufacture the product, and install it at the customer’s site. Instead of the customer paying you $20,000 up front, maybe they’re paying you $120/month. Maybe you can repeat this a few times, but sooner or later, even with customers and revenue, you won’t have the cash on hand to continue to grow. This isn’t just a hypothetical problem: SolarCity just ran into this issue and solved it by selling $227 million of installed systems to a bank. Unfortunately, your startup isn’t likely to get the same deal.

Making the Model Work

Given the difficulty of scaling companies that make physical products, is there a path forward for the leasing model for startups?

A handful of investors are working on solving this exact problem by separating equity finance from project finance. For mature technologies, Jigar Shah’s current venture, Generate Capital is actively providing funding for projects using reliable and proven but under-installed technology. This model is fundamentally different from venture capital, because it doesn’t assume the high risk and therefore doesn’t require the same rocket-ship growth. For those slightly less mature technologies, there is also hope: Rob Day at Black Coral Capital has structured a few deals where the equity for the startup has been separated from the debt for the projects and installations.

So what does this mean for your cleantech startup? On the one hand, high up-front costs are likely to remain a deterrent to customer adoption, and “no money down” will remain an attractive option. At the same time, banks will remain unlikely to lend to a risky startup without a balance sheet and investors are keenly aware of the pressures this model will put on the business. Investors like Generate and Black Coral who understand project finance for emerging (but not totally new) technology may be the only path forward. But beware: if your company is developing a next-generation battery AND trying to sell it using a no-money-down model, the leasing model just might kill your startup.

Climate Change: Deployment vs. Innovation

In Paris last week, Bill Gates and Barack Obama made headlines when they announced two new initiatives to fight climate change. The governments of 20 countries announced “Mission Innovation” where they made a commitment to double their energy research, development, and demonstration (RD&D) budgets over the next 5 years. (Bill Gates made a similar announcement earlier this summer, where he pledged to put $2 Billion of his own money to work in the sector.) The group of 28 billionaires, for their part, announced the “Breakthrough Energy Coalition” as a vehicle to invest in early-stage companies that have the power to transform the clean energy landscape.

The worldwide headline-grabbing attention that these announcements received reignited the ongoing debate over innovation vs. deployment. It prompted energy innovators, entrepreneurs, and investors alike to rehash the debate. Even Energy Secretary Ernest Moniz weighed in on the topic.

The “with us or against us” split

You could imagine a spectrum with two extreme positions. At one extreme, the deployment straw-man would say all available resources — public and private — should be spent on deploying existing technology. At the other extreme, the innovation straw-man says the existing technology isn’t cheap enough, so we shouldn’t bother installing it and all resources should be spent on next generation technologies.

To be clear – no serious voices are arguing for either of these extreme positions. This is like saying we must make a binary choice between either researching cures for cancer, or only using our current drugs to treat patients. Nevertheless, there are constraints on time and money, and there is a debate brewing about which part of the climate change problem public funds, private investment, philanthropic grantmakers, and public policy should be focusing on.

De-carbonizing the world’s electricity supply can broadly be separated into two categories: replacing existing dirty generation, and building new renewable capacity for first-time electrification. There are, of course, other sources of greenhouse gas emissions — notably transportation, manufacturing, agriculture, and changes in forest cover – which all need to be addressed.

When it comes to providing new electricity to the developing world, Jigar Shah has been making the case that all of the renewable technology we need already exists, and can be deployed more cheaply than fossil-fuels. On the heels of the recent announcement, the veteran solar developer and founder of SunEdison wrote a post on LinkedIn that was viewed by more than 10,000 people, where he made the case that in order to deploy this cleantech infrastructure, we need to divert only 25% of the $10 Trillion that would be spent on energy infrastructure anyway. As Carl Pope wrote in a 2014 response to a previous Gates-Shah back and forth, in some parts of the world, new wind and solar projects are cheaper than new fossil fuel generation. Pope and Jigar point out that for the developing world, the price difference is even greater because new distributed renewable resources can reduce the need for building the transmission infrastructure that would be required with centralized fossil fuel generation.

While the new announcements were trending on traditional media, Rob Day, an experienced cleantech venture capitalist, volleyed dozens and dozens of tweets back and forth where he asked the world to give business model innovators more credit. Rob agrees that we already have ready-to-go fundamental “deep tech” built on years of previous R&D (innovations in Materials Science, Chemical Engineering, Information Technology) and he argues that what we need most right now are creative new business models to get this technology adopted. In some cases, those new business models are creative new ways to finance the transitions. In other cases, progress comes from adding sensors, software, and services to existing technology – think smart lighting and smart energy storage.

So how did an announcement about strengthening R&D rekindle a debate that pits climate change fighters against each other? Where did this debate come from, and how did it get so politicized?

The Debate

First, let’s be clear about what this debate is not about:

The debate is not about whether technology R&D is important (no serious voices are claiming that it isn’t). Neither is the debate about whether government policy plays a role in deployment (both sides agree it does).

It is important to consider the pipeline of technology development. Fundamental research leads to new discoveries. Sometimes, a market need arises for those discoveries. In some cases, this “market pull” happens 20 years or more after the initial discovery. Then, a new wave of research and development is needed to bring that discovery and the manufacturing processes to create it to commercial scale. Next, a different sort of innovation is needed to demonstrate, market, and sell the new innovation. As the technology is deployed, manufacturing experience, the strengthening of supply chains, and R&D investments by the manufacturers all help bring the costs down further.

So, since all of these pieces are necessary for widespread adoption, why would R&D stand in opposition to deployment?

The Innovators

In the mid 1990s, scientists, engineers, and policy makers were looking at the magnitude of the climate challenge and realized that the problem was much bigger than people understood. Alex Trembath has written a detailed history of these voices, that you should definitely read here and here.

These thinkers made the case that new energy consumers in the developing world would demand electricity much faster than energy efficiency upgrades could keep up. By being more efficient, we would not break even, much less reduce our emissions. New technology would be needed to decarbonize the world affordably. Note here that none of these innovation advocates suggested that we shouldn’t deploy existing solutions, and that’s an important distinction.

Unfortunately, these thinkers didn’t quite get as much press as Bjørn Lomborg, a Danish political scientist.

The Skeptical, but wrong, Environmentalist

In 2001 Bjørn Lomborg, a Danish political scientist, published the English-language version of his book, The Skeptical Environmentalist. Lomborg is clearly passionate about saving the planet, and especially about looking out for the global poor. His book drew inspiration from other sources, but, perhaps because of the recent Kyoto Protocols (COP3), it garnered more attention than those prior volumes. The publication of this book was met with a healthy dose of praise from newspapers including the New York Times, the Washington Post, the Wall Street Journal, and the Economist. The book was welcome news to those who didn’t want another global threat to worry about. It’s just too bad so much of it wasn’t true.

Publications with a higher level of scientific rigor, including Nature and Science, wrote harsher reviews, accusing the work of selectively including references and even of falsifying data in order to deliberately mislead readers. Scientific American published several rebuttals by prominent scientists, followed by Lomberg’s response, and another set of rebuttals to the response.

In one chapter, Lomborg suggested that while he believed that climate change was real, he doubted the extreme temperature predictions, and claimed that his own cost-benefit analysis indicated that it was not the most pressing problem facing the world compared to poverty, disease, and global development. Instead of deploying existing clean technology, resources should be spent on directly fighting disease and lifting people out of poverty, and some funds should be spent on R&D for future clean energy solutions. The scientists who reviewed this section found that he failed to include the most recent scientific studies, and misinterpreted others. His calculations led him to the conclusion that the climate worries weren’t as large as others feared.

Environmental scientists worried that Lomberg’s book had the appearance of legitimacy, and therefore would be used as evidence for those who opposed any action on climate change. Clean energy advocates, meanwhile, worried that if the world perceived that next year’s clean tech would be cheaper and better than what’s available this year, they would perpetually hold off on making the conversion. As Jesse Jenkins summarized it: in order to say that R&D is needed, you have to admit that the current technology can be improved. That is not the same as saying today’s technology isn’t worthwhile.

The richest man in the world enters the fray

In 2011 Bill Gates gave a talk in which he said that the climate benefits of energy efficiency technologies would be wiped out by increasing fossil-based energy use in the developing world. In the summer of 2014, Gates weighed in on his blog by posting two of Lomborg’s videos. In those two videos, Lomborg starts with reasonable arguments: the world’s poor need access to energy in order to lift themselves out of poverty and that indoor cooking fires are terrible for human health. No one could disagree. But then he continues by making the unsubstantiated case that these two problems can only be solved by fossil fuels, ignoring the many cases where renewables can cost-effectively play a role.

In discussing these videos, Bill Gates made four proclamations:

  • The rich world should not put emissions constraints on the developing world if that would harm their ability to fight poverty.
  • Developing countries should be given access to the cheapest sources of energy they can find, in order to accelerate their development.
  • The poor countries wouldn’t contribute enough emissions, even with dirty energy, to make a significant difference.
  • To fight climate change, rich countries should spend more money on R&D for technologies that will make clean energy more affordable to everyone.

On their face, all of these statements are reasonable, and I believe even the staunchest deployment-only advocate would agree with all four. The problem here is that, even though his blog post says he “doesn’t agree with everything Lomberg says,” Gates both tacitly and explicitly agrees that the clean alternatives are too expensive. Gates also did a disservice to the clean energy industry and the global poor by failing to acknowledge the many cases where clean alternatives are both cheaper and more reliable. This is perhaps a little surprising, considering that during Gates’ tenure as Chairman at Microsoft, the company signed huge wind power deals.

Innovating vs. deploying in established markets

Now, on to the second part of the debate – innovating in established markets.

Rob Day has made a strong case for business model innovation as the economic driver for deploying existing solutions. The question has been: with all of the clean energy solutions available today, what can we do to get them to market quicker, and how do we ensure that there are businesses focused on this?

These technologies fall into two categories. First, technologies that are cost-effective over the long term, but require substantial up-front costs or long payback periods. For example, installing rooftop solar may save thousands of dollars a year, but might cost up to $20,000 to install and have a payback period of 10 years. This is the problem SolarCity is solving, by providing home owners with the benefits of solar on their rooftops without the burden of paying out of pocket for installation or maintenance.

The second category is technologies that have a smaller financial benefit, but where additional value streams can rapidly increase uptake. Simply swapping out inefficient lighting for LEDs, for instance, has a payback period of about 2 years for industrial customers. Manufacturing and selling commodity LEDs is a tough business, and not likely to attract risk-tolerant early-stage capital. One of our portfolio companies, Igor, connects the lighting to sensors and control systems, allowing the system to automatically dim or shut off the lights in response to ambient conditions.

The path forward

Suggesting that more R&D is needed is not the same as saying today’s solutions aren’t worth deploying. There is little question that next year’s automobiles will be always be better than the ones on the lot today, but that doesn’t stop people from buying. At the same time, anyone who is active in the clean energy space has to remain vigilant against opinions about the relative costs and benefits that aren’t supported by robust data.

Most people who support R&D for the next generation of technologies are not doing so in order to block or delay the transition away from fossil fuels. Instead of telling Bill Gates he’s wrong to invest his money in next generation technology, our community needs to show him that by also focusing on deploying current technology, we can save the lives of those he is trying to help.

So what’s the path forward? First, there is no question that regulatory barriers to deploying existing, cost competitive technology should be broken down. Second, government policies around the world should continue to support the roll-out of deployment-ready technologies to help those solutions come down the learning curve.

Finally, research and development for the next generations of energy technologies and climate mitigation technologies must be supported. But R&D alone won’t be enough. Companies formed around new materials, chemicals, and manufacturing processes face immense hurdles on the road to commercialization. These companies are not a good fit for traditional risk capitalists who need substantial returns and short investment timelines. Successful venture-backed companies must multiply the investment 10- to 100-fold in 3-5 years. New models are arising to help fill this gap, in the public, private, and non-profit sector. In another piece well worth reading, Teryn Norris has done a great job of summarizing these models here.

The hope is that Bill Gates and the rest of his coalition will find ways to support and commercialize the new innovations that we need to win the fight against climate change, and will make money while doing so.

This will be a long battle, but it is one that the entire cleantech community needs to fight together.

A World without Patents?

“Set Innovation Free” read the cover of the economist last month. You can skip the fluffy leader, and dive straight into the main article, or better still, the working paper by economists at the St. Louis Federal Reserve Bank upon which most of the article is based.  The article posited that we would have more innovation without patents, that patents don’t encourage the behavior they are designed to encourage, and that gradual rollback of patent laws — leading to abolition — would be good policy.

As an intellectual property advisor and a patent applicant, I want strong protection for my ideas and my clients. At the same time, as a former patent examiner, I’ll be the first to admit that our patent system isn’t perfect (one small example — average time between application and the first response from the office for software patents: 21 months). But the case for doing away with the system altogether isn’t often made. To understand the arguments for and against, let’s start by looking at why we have patents in the first place.

What good are patents?

Put simply – patents let an inventor block others from using or selling the invention (granting a monopoly), in exchange for sharing information about the invention. The inventor can, of course, let people use the invention and charge a licensing fee, or give it away for free (like Tesla did for its charging technology).

Patents promote innovation in two ways. First, patents encourage an inventor to invest in a new innovation because it gives exclusive rights to sell that invention. Second, the patent discloses how the invention was made, giving everyone an information advantage and speeding up the next round of invention.

There are three primary arguments in favor of a strong patent system:

1) Open-source invention. Because a patent describes the invention in detail, new inventors can stand on the shoulders of those who came before and move the field along quicker.
2) Incentivize investment in new invention at big companies. Large public companies are answerable to their shareholders. If they can’t reap the financial rewards of an innovation, they will not invest in innovation.
3) Encourage new players to enter a market. If a small startup can’t protect its new invention, an existing well-capitalized business with existing supply chains, manufacturing, distribution, and a strong brand can copy the idea and crush the startup.

There’s a fourth argument that doesn’t come up quite as often, but is just as important:

4) Countries with stronger IP attract more investment. When companies choose where to manufacture, they prefer to keep critical patent-protected inventions in countries that guarantee them their intellectual property rights.

The case against patents

Given the strong case for patents, why would the economist suggest we should get rid of them?

The arguments against our system can be summed up in three points:

1) Patents do not actually lead to more innovation.

2) There are lots of negative side effects.

3) Any patent system will evolve over time to be better for established players and worse for new entrants.

Less innovation with strong patents?

Recall that patents are supposed to increase the incentives to innovate in the first place, and then make subsequent innovations (improvements) easier.

The article raises a few interesting examples of studies indicating that patents don’t always lead to more invention. For example, a study of inventions at international fairs in the 1800s  showed that the rate of invention was no different between countries with and without patent systems. In the 1970s, new rules that expanded the ability to patent crops didn’t lead to more R&D or increases in wheat yields.

After that, the examples get a bit weaker. According to the authors, prior to a new rule allowing German companies to patent new drugs, they invented more new drugs than British companies. Unfortunately, they don’t provide any details about what happened after the rule was changed, nor do they compare Germany to other countries.

The article is weakest when it quotes an article from 1851:

Most of the wonders of the modern age, from mule-spinning to railways, steamships to gas lamps, seemed to have emerged without the help of patents. If the industrial revolution didn’t need them, why have them at all?”

Nevermind that all of these inventions were patented: the spinning mule was patented in the 1700s, and the first automatic mule was patented in 1825, gas lamps were patented in the late 1700s/early 1800s, rail designs were patented, and steam engines were patented in the 1600s. It is easy to say patents “weren’t needed”, but since each of these components of the industrial revolution was patented, the burden of proof is on the authors to show that these would have been invented anyway.

As to whether patents encourage follow-on innovation, the authors of the paper also state that patents are written using confusing legal language, and therefore do not accomplish the goal of spreading knowledge of the invention. More interestingly, the authors state that companies instruct their engineers not to study existing patents, in order to protect them from claims of infringement. (If they didn’t know about the patent, they couldn’t willfully infringe.)

These authors, and the economist article, have provided some evidence that innovation isn’t increased by patents, but a few examples from the 1800s and anecdotal evidence about what happens at a few large companies leaves a great deal of further study to be desired.

Side effects

Both the article and the St. Louis paper make the case that the many negative side-effects of our current patent system are enough to warrant some major changes.

The article suggests that a patent system:

  • restrains free trade
  • restrains competition
  • encourages fraud
  • encourages rent-seeking
  • creates arguments among inventors
  • creates lawsuits
  • allows patent trolls to stop inventions (block competition)
  • enables patent trolls to “appropriate the fruits of the inventions of others”
  • does not give security to “really good inventors”

There is ample evidence for each of these points, which leads many to agree that the system simply needs to be reformed to prevent these side effects. That brings us to the final point:

Evolution of the system

The article maintains that any patent system will devolve to encourage rent-seeking behavior as large, established players seek to protect their market share. These companies, they say, will use their leverage to lobby for rule changes that favor them, and that they can spend their substantial wealth to buy defensive patents and bring legal action (or the threat of legal action) to block smaller entrant firms.

The economist’s patent-free paradise

So what would the world look like without patents? The authors acknowledge that in some fields, like drug discovery, a patent system does encourage R&D spending. A better model, they suggest, would be a prize competition where firms who develop a new drug receive significant funding from the government. The drugs could then be manufactured by anyone, reducing the sale price of the drug, thereby reducing Medicare and Medicaid payments. Those welfare savings would more than pay for the prizes. While this sounds like a reasonable approach, I’m afraid it would fail in the details. While the government could announce a list of highly sought-after life-saving cures (say: malaria, HIV, specific cancers), how many prizes would the government set aside for drugs that improve the quality of life, like those that slow the progression of dementia, or treatments that aren’t covered by Medicare and Medicaid.

Closing Thought: Are patents different from copyrights?

Many of the same arguments for and against a patent system can be made for copyright protection as well. If the Economist is serious, maybe the newspaper should give up copyright protection on its content?

What happened to US Solar? Part 1: OptiSolar

Today the headlines are filled with great stories about successful solar companies: vivint, SolarCity, SunEdison. But what about all the news stories about all the US solar companies that went belly-up over the last decade?

Some have cited bets on the wrong technologies: CIGS, ink-based cells, thin film cells. Some have said that it was a short-sighted gamble on an ever-increasing price of bulk polycrystalline silicon, while some blame Chinese manufacturing — cheap capital, easy permitting, and established supply chains helped vastly oversupplying the market — and some further suggest that China was dumping panels (that is, selling them for less than they really cost in order to corner the market and push out other suppliers.) Still others have put the blame on venture capital: that the VC model wasn’t right for energy, or that the investors put in too much money in too many companies, or that there wasn’t enough venture capital to get companies through to an exit. Now that we’re a few years out of the big “cleantech bubble,” I’ll be diving in to a few of those companies on a case-by-case basis.

This is part one of a set of case studies where I dive in to what happened to a few of the colossal failures. The blog post is informed by a number of great news articles and press releases, which you can find at the bottom of the page.

Part 1: OptiSolar

All we need is scale

In 2004, two engineers at Hewlett-Packard predicted that the future of solar was “super-large-scale manufacturing”. The idea was simple, set up “solar cities,” where every piece of the supply-chain is co-located, own every piece of that supply chain (even the solar farms themselves), and, most importantly, doing it at incredibly large scales. That was the key — the cost of solar could only be price competitive at a massive scale. Marvin Keshner and Rajiv Arya, the two engineers, floated the idea to their bosses at HP, but management wasn’t interested.

Instead, in 2005, they took over the patents and founded OptiSolar. They believed that costs could come down “without the need for any significant new innovation. It [low cost] comes entirely from the design of a very large, dedicated and optimized factory, the design of manufacturing equipment for a very large factory and the cost savings resulting from operating at such a large manufacturing scale.”

OptiSolar suprised the world of renewable energy in April 2007 announcing that they would install North America’s largest solar farm, 40MW, in Sarnia, Canada, using panels produced in Silicon Valley.

Rising silicon prices

OptiSolar had something big going for it: the price of silicon was skyrocketing. The silicon used for solar panels had historically come cheap — the unwanted byproducts of the computer chip industry. As the demand for solar panels grew, so did the price. Between 2005 and 2007 that price had more than tripled. Contenders in the solar arena were busy inventing new solar cells based on other materials or ways to use considerably less silicon.

OptiSolar reduced costs by using a very thin film of amorphous silicon. Press releases touted their ability to recycle silane gas, the raw material source of the silicon in their process. Even though thin-film technologies were known to have lower efficiencies compared to other technologies, the founders were confident that their vertically integrated business model would mean they could stay competitive, once they could manufacture at scale.

The Sarnia project

When they announced the contract to develop the Sarnia solar farm, their manufacturing plant was still under construction in Hayward, California. Canada was an obvious first choice because OptiSolar could take advantage of generous feed-in-tariffs, essentially subsidies for renewable power generation. The Ontario Power Authority agreed to a 20-year deal to buy the power for C$0.42/kWh, (about US$0.46 /kWh at the time, and has hovered between US$0.40 and US$0.50 since). At the time, average retail prices for electricity in the US were US$0.09/kWh. The project was planned in four 10MW phases, and OptiSolar said at the time that a standard 10 MW installation at the time would cost $C 70-80 million. At that rate, the full project would cost around $320 million. A separate estimate pegged the cost of manufacturing the panels at $300 million — not counting the installation. Sources said that by the end of the year, the company had raised between $35 and $65 million. There were questions about whether the technology could perform as promised — at that point, OptiSolar had yet to demonstrate a solar module or solar panel that used their thin film technology.

More projects, more money, and a shiny factory

Yet to deliver on the promises of the Sarnia plant, in early 2008 the agreement for the Canadian project was nevertheless upgraded from 40 to 50 MW, and at the end of January, the firm raised over $38 million. This was followed in April by a flurry of new announcements: the planned Canada project had grown to 60 MW, another 140 MW of deals elsewhere in Canada had been signed, and another $132 million of new funds had been raised. In July they raised another $77.8 million, bringing the total funding to over $300 million, all of which was equity investment, according to Alan Bernheimer, the Vice President for corporate communications.

This may have been the most prescient thing OptiSolar did. The founders realized earlier than most others that the “high capital requirements would exceed the capabilities and sensibilities of Sand Hill Road.”[1] Instead of raising funds from typical venture capital, the investors were oil and gas private equity firms, mostly based in Canada. The private equity investors presumably had the patience for energy investments that VCs did not.

In conjunction with announcing the latest fundraising, OptiSolar announced that a new 550 MW solar farm was in the works in San Luis Obispo County, California. In August, the company revealed that they had won a competitive contract from Pacific Gas & Electric to purchase the power from this field. The project, dubbed the Topaz Solar Farm had an estimated cost of $1 billion.

Construction of Topaz wasn’t slated to begin until 2010, but because construction in Canada had yet to begin and the company had still not presented a product, some questioned how an unproven startup with an untested technology could land such large deals. Bernheimer attributed their success in securing the contracts to their competitive advantages in automated manufacturing and vertical integration — neither of which were up and running. By October of 2008, OptiSolar stated that the company wasn’t worried about the credit crunch, claimed that the first 10 MW phase of generation at the solar farm in Sarnia would be up and running by the end of 2008, and vice president Peter Carrie expected the rest of the project to be completed in 2009.

Around the same time, the company announced a new million-square-foot factory in Sacramento, able to produce 600 MW worth of panels per year. To lure OptiSolar to the former site of McClellan Air Force Base, the county had offered $20 million in tax rebates. In November, Governor Schwarzenegger and TV crews from 60 minutes visited the factory and used it as a backdrop to sign a new executive order supporting the renewable energy industry.

The downturn

Three days after the Governor’s visit, the company suspended work on the factory until the following year. Asked why, Bernheimer explained that in “tough economic times, you husband your resources.” The company — that hadn’t been worried about the recession a few months before — said it needed another $200 million to continue expanding the new factory, but the current crop of investors either weren’t willing or weren’t able to pile on more cash.

The bad news piled on quickly. In January of 2009, OptiSolar announced that it would lay off half its employees, 105 at the plant in Sacramento, and 185 in Hayward. Construction of the million-foot plant in Sacramento would be delayed until at least the second half of the year. The company claimed to be continuing production of modules in Hayward for the Canada project, but had pushed the target delivery date by a full year. At this point, Bernheimer claimed that the first 10 MW phase would be complete by the end of 2009. The company announced that it would apply for a loan guarantee from the Loan Programs Office at the Department of Energy.

If there was a bright spot for the company, it was the first announcement of product delivery: 1MW of panels had been installed in Sarnia. This is compared to the 2,000 MW OptiSolar had commited to across North America. Another glimmer of hope came in February, when the PG&E deal was formally approved by the utility commission in California, and the company filed the paperwork to request a $300 million loan guarantee.

Refocus on manufacturing or a return to R&D?

In early March, OptiSolar’s rival First Solar announced that it would take over OptiSolar’s project pipeline – all of the agreements to install and operate solar farms. This amounted to approximately 2 GW of capacity — including the Sarnia and 550 MW Topaz projects. In exchange, OptiSolar would receive $400 million worth of shares in First Solar. At the time, there was speculation that this was an effort to refocus the company’s efforts on manufacturing, but later interviews with one of the executives indicated that the private equity investors pressured the sale in order to recoup their $322 million investment.

Whatever the reason, OptiSolar was now only a manufacturing company. Technology analysts, though, weren’t convinced that the company’s numbers added up. Doubts were raised about efficiencies, outputs, and production capacity. Silicon raw-material prices had peaked in 2008 at $450/kg, 6 times higher than they were in 2005. By 2009, the price had crashed to $100/kg. The cost-savings of the thin-film technology wasn’t as meaningful anymore. As for the modules, the efficiency was estimated at 5 percent to 5.5 percent by GTM Research, while at the time the industry standard was 6.5 percent. The company’s claim all along was that even though the thin-film technology would be less efficient, the lower costs of manufacturing at scale would make up the difference. Outside estimates indicated that, even with these scaling effects, efficiencies in the 9 – 10 percent range would be needed to stay price-competitive.

The projects that survived

Later that same month, OptiSolar announced that it would stop manufacturing and lay off most of the remainder of its staff. The planned facility in Sacramento would lose 58 staff, and 142 at the original factory in Hayward would lose their jobs. Bernheimer said at the time that production was ready to start. But a buyer would need “resources, cash flow, and the ability to invest in research and development” in order to get the factory up and running. His words may have revealed the actual maturity of the technology. Meanwhile, analysts predicted that any company relying on the manufacture-at-scale model would “burn through its cash before it can start to ship in volume for a decent revenue stream, fail to find more backing, and be forced to pull the plug.”[2]

The Sarnia project, at least, was mostly spared. After a few regulatory hurdles, First Solar, now the developer of all of OptiSolar’s projects, was in the process of removing the 2.5 MW of installed OptiSolar technology and installing up to 80 MW of First Solar’s cadmium telluride technology.

Two years later, in 2011, construction began on Topaz, the biggest First Solar project acquired from OptiSolar. First Solar applied for a loan guarantee for the project, but was unable to secure financing. They were saved by Warren Buffet — a month later after construction began, the project was bought by Berkshire Hathaway’s MidAmerican Energy Holdings. Construction was planned to be complete by 2015, though the total projected costs had doubled to $2 billion.

Pulling the plug, and OptiSolar’s legacy

In July of 2009, the Canadian company EPOD Solar came forward to buy OptiSolar’s intellectual property and manufacturing capacity for $260 million in stock. EPOD Solar was really Allora Minerals, a Canadian mining company that had bought the assets, and name, of EPOD Solar. The OptiSolar assets up for sale included the Hayward facility, which was revealed to have only 15 MW of production capacity. OptiSolar had invested about $310 million in the manufacturing capacity. The million-square-foot factory that Schwarzenegger had visited was also included in the purchase, and promptly listed for sale on a real estate website.

OptiSolar’s projects were in good hands with First Solar, and OptiSolar’s investors were at least reasonably happy that the $400 million in First Solar stock pay back the investors (who had contributed $322 million). EPOD Solar, the buyer of OptiSolar’s equipment and IP had changed hands and was now NovaSolar, a subsidiary of a Hong Kong company.

NovaSolar, the new upstart, had a clear vision of “utility-scale power plants that will essentially undercut any other vendor on the planet” and solar modules based on existing technology, once they reach high-volume production.

One of the founders of the company was Marvin Keshner. If that name sounds familiar, it should. Keshner was the original author of the HP report, and a founder of OptiSolar. The other founders were also former OptiSolar alums.

NovaSolar secured investment from Asia, leased 65,000 square feet of research and development space in Fremont, California, and started building a 500,000 square-foot factory in China capable of producing 250 MW of panels per year. According to COO Darien Spencer, the business model remained the same as before: the “end product is building power plants and selling the power to utilities and utilizing your own product [the solar panels].” Apparently, the hype was the same as before, too, “it’s a great opportunity because as you reach grid parity, you have unlimited market potential.” The technology was also the same, with a few years of R&D improvement. The company claimed that efficiencies were now in the 8-9 percent range, making it better than when OptiSolar failed two years earlier, but not enough to keep up with improvements in competing technologies. The difference this time? Access to cheap finance in Asia. “Money is easier to borrow and factories easier to build.”

Apparently, cheap financing and easy construction permitting weren’t enough to fix the problems. By February of 2012, the San Francisco Business Journal reported that NovaSolar had furloughed 52 of 60 employees, and that the remaining eight had been unpaid for months. Construction in Fremont and China had been tabled, with contractors claiming $1 million in unpaid work so far.

In June of 2012, NovaSolar filed for bankruptcy.

So what really happened?

OptiSolar, and NovaSolar after it, failed because it underestimated the difficulty of taking an unproven technology to massive scale, and the speed at which innovation would occur in other companies. Other factors certainly played a role, too. OptiSolar was not helped by the decline in bulk silicon prices, but even if silicon prices had remained high, other non-silicon competitors (like First Solar) would have won the day. Even if the company had found investors or lenders and finished the 600MW factory, the panels would likely not have been cost competitive and the company would have trouble winning new development contracts.

Unfortunatley, OptiSolar wasn’t the first or last company to try to prematurely scale a new technology, but I’ll leave that for future posts.

The good news? Topaz, the world’s largest solar farm, came online in late 2014 and was joined by another 550 MW First Solar plant in early 2015.



References: (Jan 2008) (4/29/2008) (8/15/2008) (8/18/2008) (12/05/2008) (1/18/2009) (03/19/2009) (3/20/2009) (4/13/2009) (7/21/2009) (7/22/2009) (7/22/2009) (Nov 2009) (2/24/2012)